Presentation 2015-03-16
Further Speeding Up and Solution Quality Improvement of Singularity Stairs Following
Seiya SATOH, Ryohei NAKANO,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) In a search space of a multilayer perceptron (MLP), there exists singular regions where any point is I-O equivalent to the optimal solution of an MLP having one less hidden units. Singularity Stairs Following (SSF) utilizes singular regions as initial points for search and systematically finds excellent solutions increasing the number of hidden units one by one. Since the number of search routes in SSF rapidly increases as the number of hidden units gets larger, search pruning has been proposed to deal with the problem. In this paper, we try to make SSF even faster by limiting the number of search routes without deteriorating solution quality.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) multilayer perceptron / learning method / singular region / reducibility mapping / search pruning
Paper # MBE2014-168,NC2014-119
Date of Issue

Conference Information
Committee NC
Conference Date 2015/3/9(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Neurocomputing (NC)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Further Speeding Up and Solution Quality Improvement of Singularity Stairs Following
Sub Title (in English)
Keyword(1) multilayer perceptron
Keyword(2) learning method
Keyword(3) singular region
Keyword(4) reducibility mapping
Keyword(5) search pruning
1st Author's Name Seiya SATOH
1st Author's Affiliation Chubu University()
2nd Author's Name Ryohei NAKANO
2nd Author's Affiliation Chubu University
Date 2015-03-16
Paper # MBE2014-168,NC2014-119
Volume (vol) vol.114
Number (no) 515
Page pp.pp.-
#Pages 6
Date of Issue